The Harnack theorem for elliptic systems degenerate at an inner point (Q1280869)

In a bounded domain \(D\subset \mathbb{R}^n\) with boundary \(S\), we consider the system \[ Lu:= \sum^n_{i,j=1} A_{ij}(x) u_{x_ix_j}+ \sum^n_{i=1} B_i(x) u_{x_i}+ C(x)u= 0,\tag{1} \] where \(A_{ij}= \text{diag}(a^{(1)}_{ij},\dots, a^{(N)}_{ij})\), \(B_i= \text{diag}(b^{(1)}_i,\dots, b^{(N)}_i)\) and \(C= (c_{ks})^N_{k,s=1}\) are real matrices continuous in \(\overline D= D\cup S\), \(u= (u_1,\dots, u_N)\) is an unknown vector-column. In the case \(N=1\), the Harnack theorem applies. Namely, if an equation is uniformly elliptic in \(D\) and \(C(x)\leq 0\), then the uniform convergence on the boundary \(S\) of the sequence \(\{u_k\}^\infty_{k=1}\) of its solutions \(u_k\in C^2(D)\cap C(\overline D)\) implies the uniform convergence of this sequence to the solution \(u\in C^2(D)\cap (\overline D)\) of (1). The proof of the Harnack theorem is based upon the Hopf maximum principle, which holds under the assumption \(C(x)\leq 0\). We prove an analogue of the Harnack theorem for system (1) under several assumptions.